-
-
Notifications
You must be signed in to change notification settings - Fork 254
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
make Java install resilient #878
Comments
No, @mstormi I was querying the Latest binary and that is what Azul returns: compare these links: |
The builds started failing about two days ago. We could have caught it but I was busy trying to get RL things done and wasn't paying enough attention to make the connection. What you did to fix it should suffice in all cases except when Azul is in the middle of uploading their latest builds and the |
Same for me. But relying on that is a broken concept. Maintainers cannot be on duty 24x7, 365 days a year. Hence my focus on a strategic approach to mitigate the need for fast reaction.
Ok, then skip looking for a final fix and move on to the strategic approach I raised. |
My one worry about keeping local copies is that then we can run into licensing issues with the packages that we include. |
For Zulu Community the Azul website says you can "download and use without restrictions". Adopt license is Apache 2.0 https://github.com/AdoptOpenJDK/openjdk-build/blob/master/LICENSE which should be fine as well (same as for the Balena sources). If anyone finds information to deviate, please point me there. |
For the sake of tracking, no as it's completely different issues. |
I would but right now I'm more or less in hurry up and wait mode for a fix from Azul. |
FYI The latest Raspbian (lite) image that is available for download includes Oracle JDK and other packages that require licensing. For their handling see here |
Do we need to worry about the licensing? |
what do you mean ? because we in turn include Raspbian ? |
Yeah |
I think we should leave a reference on the download page like the Raspbian guys did. |
Would you say we're done with this ? |
Yes, until such a time that Azul gets back to me with a proper fix we are done with this. |
Additonally fix Java installation output. Fixes openhab#878 Fixes openhab#879 Signed-off-by: Ethan Dye <[email protected]>
Additionally fix Java installation output. Fixes openhab#878 Fixes openhab#879 Signed-off-by: Ethan Dye <[email protected]>
Additionally fix Java installation output. Fixes openhab#878 Fixes openhab#879 Signed-off-by: Ethan Dye <[email protected]>
Additionally fix Java installation output. Fixes #878 Fixes #879 Signed-off-by: Ethan Dye <[email protected]>
Following up on #874, there is a need to make (first & foremost but not exclusively) Java install more resilient.
Need to consider both a short term and a strategic approach in parallel.
On short term, we need to evaluate the list of available packages and select hf ones.
@ecdye I didn't have the time to dissect this in detail at 2am so may be wrong, but unlike what you wrote in #874 the URL OHian requested returned MORE than one package and your code seemed to select the last out of those. So it was bad luck that that was the sf one. Double-check that please
and eventually come up with a (non-hot) fix.
On strategic, see #655. I stand by my opinion to keep copies of latest packages known to be good
(= have a copy in openhabian repo). That we should do at least for Azul and Adopt Java.
[NOTE: we may need to have several copies for different HW architectures]
Mind #692: given AC's 'tie break' that took place there, I'd like to get this implemented asap.
It means we'll have a new "stable" branch in addition to today's master branch.
What it means for this issue here is that changed Java install code should evaluate the branch first and if in 'stable', it should download the known good package off the openHABian repo.
Also mind the connection to #825 (offline install), our image would need to contain a copy as well for that to work. So even with Inet available, on the system is the first place place the code should look for.
If in master it should continue to retrieve the package directly from the source, BUT it should be extended to fall back to the repo cache if the install routine can determine that downloading from source fails or is bad somehow [it obviously cannot detect all instances of 'bad' such as the aarch32sf one, but it can detect some of them].
PS: we should duplicate this approach for other external sources once it has proven to work
PPS: there would have been a chance to spot #874 early: actually all test builds made since the Azul change were failing.
The text was updated successfully, but these errors were encountered: